Bug 761812 (GLUSTER-80)

[Migrated from RT] - ticket 814 [http://support.gluster.com/rt/Ticket/Display.html?id=814]

Wed Mar 11 05:08:56 2009  	 guru - Ticket created  	 	 
Version: glusterfs 2.0.0 pre30

..
87: volume ioc
88: type performance/io-cache
89: subvolumes unify
90: option cache-size 500MB
91: option cache-timeout 3600
92: end-volume
..
2009-03-11 15:35:10 W [xlator.c:167:_volume_option_value_validate] ioc:
'3600' in 'option cache-timeout 3600' is out of range [0 - 60]
2009-03-11 15:35:10 N [glusterfsd.c:1134:main] glusterfs: Successfully
started

--------------------------------------------------------------------------------
#   	Wed Mar 11 07:59:17 2009 	amar - Correspondence added

Subject: 	Re: [support.gluster.com #814] [ glusterfs 2.0.0 pre30 ] - Glusterfs does not error out upon out of range option value
Date: 	Wed, 11 Mar 2009 07:59:15 -0700
To: 	rt
From: 	Amar Tumballi <amar>
Download (untitled) [text/plain 684b]
This is intentional. We don't want to error out for range check. The idea
was to give an idea to users about the possible range, but even the out of
ranges should work fine. Also, for few options, we *think* a range suits it
best, but are not sure if thats the best. hence we are not erroring out, but
giving users to try different values.

Regards,

On Wed, Mar 11, 2009 at 5:08 AM, Gururaj K via RT <rt> wrote:

>
>
> ..
> 2009-03-11 15:35:10 W [xlator.c:167:_volume_option_value_validate] ioc:
> '3600' in 'option cache-timeout 3600' is out of range [0 - 60]
> 2009-03-11 15:35:10 N [glusterfsd.c:1134:main] glusterfs: Successfully
> started
>
>
>


-- 
Amar Tumballi

--------------------------------------------------------------------------------
#   	Thu Mar 12 00:21:03 2009 	vikas - Correspondence added

On Wed Mar 11 07:59:17 2009, amar wrote:
> This is intentional. We don't want to error out for range check. The idea
> was to give an idea to users about the possible range, but even the out of
> ranges should work fine. Also, for few options, we *think* a range
suits it
> best, but are not sure if thats the best. hence we are not erroring
out, but
> giving users to try different values.

That does not make sense. If we want to 'recommend' certain values to
users, the proper place to do it is in the documentation.

When we are _validating_, then the minimum and maximum values should be
strictly enforced. If this validation is not strict, then some
translators will have to do validation on their own (for example,
io-threads might want that number of threads be atleast 1), thus
defeating the whole purpose of xlator_options.

--------------------------------------------------------------------------------
#   	Thu Apr 23 17:49:13 2009 	gowda - Correspondence added
what is the status of this error report?

-- 
gowda